Optimizing Quinoline Derivatives for ABCB1 Inhibition: A Machine Learning Approach to Combat Multidrug Resistance in Cancer

Authors

  • Mouad Lahyaoui Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Riham Sghyar Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Yousra Seqqat Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Fouad Ouazzani Chahdi Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Ahmed Mazzah University of Lille, CNRS, USR 3290, MSAP, Miniaturization for Synthesis, Analysis and Proteomics, Lille, France.
  • Amal Haoudi Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Taoufiq Saffaj Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.
  • Youssef Kandri Rodi Laboratory of Applied Organic Chemistry, Faculty of Science and Technology, Sidi Mohamed Ben Abdellah University, USMBA, P.O. Box 2626, Fez, Morocco.

DOI:

https://doi.org/10.9734/bpi/cicms/v9/8483E

Keywords:

QSAR, quinoline, PCA, machine learning, deep learning, molecular docking

Abstract

A vast array of human tumors contain multidrug resistance (MDR) proteins linked to the ATP-binding cassette family, which lead to treatment failure. One of the mechanisms of multiple drug resistance is the overexpression of efflux pumps, like ABCB1. In order to predict the inhibitory biological activity towards ABCB1, the goal of this paper is to develop a robust quantitative structure-activity relationship (QSAR) model that best describes the correlation between the activity and the molecular structures. Using various linear and non-linear machine learning (ML) regression techniques, such as k-nearest neighbors (KNN), decision trees (DT), back propagation neural networks (BPNN), and gradient boosting-based (GB) methods, a series of quinoline derivatives of eighteen compounds were examined in this regard. Their goal is to identify the source of these compounds' activity in order to create new quinoline derivatives that have a stronger effect on ABCB1. A total of sixteen machine learning (ML) predictive models were created using varying numbers of 2D and 3D descriptors. The statistical metrics root mean square error (RMSE) and coefficient of determination (R2) were used to assess the models. With one descriptor, represented by R2 and RMSE of 95% and 0.283, respectively, a GB-based model, specifically catboost, achieved the highest predictive quality among all developed models. The outward-facing p-glycoprotein (6C0V) was the target crystal structure for molecular docking studies, and the results showed strong binding affinities via both hydrophobic and H-bond interactions with the relevant compounds. At -9.22 kcal/mol, the 17 has the highest binding energy. As a result, it is possible that structure 17 will prove to be a useful potential lead structure for the synthesis and design of more effective P-glycoprotein inhibitors that can be combined with anti-cancer medications to manage cancer multidrug resistance.

Downloads

Published

2024-05-01

How to Cite

Mouad Lahyaoui, Riham Sghyar, Yousra Seqqat, Fouad Ouazzani Chahdi, Ahmed Mazzah, Amal Haoudi, … Youssef Kandri Rodi. (2024). Optimizing Quinoline Derivatives for ABCB1 Inhibition: A Machine Learning Approach to Combat Multidrug Resistance in Cancer. Current Innovations in Chemical and Materials Sciences Vol. 9, 157–184. https://doi.org/10.9734/bpi/cicms/v9/8483E

Issue

Current Innovations in Chemical and Materials Sciences Vol. 9

Section

Chapters